Patrick Pribil1, Stephen Tate1 and Kristina Jurcic2
1SCIEX, Canada; 2Bioinformatics Solutions Inc., Canada
ZT Scan DIA on the ZenoTOF 7600+ system improves protein sequence coverage and reduces protein identification false discovery rate
Key features of high-depth proteomics using ZT Scan DIA on the ZenoTOF 7600+ system
- Easy method setup: Minimal optimization is required for ZT Scan DIA setup, with pre-built methods designed for a range of desired chromatography regimes
- Increased coverage: Protein sequence coverage is improved by as much as 7-fold in highly complex digests using ZT Scan DIA, enabling users to gain deeper insight into their biological research
- Maximize specificity: ZT Scan DIA reduces the false discovery rate of protein identification through greater specificity achieved using the scanning quadrupole dimension
Figure 1. Overlap of identified proteins (A, C) and peptides (B, D) from Lysate Mix 1 (A, B) or Lysate Mix 2 (C, D). Lysate Mix 1 is a mixture of protein digests from human, yeast, mouse and Chinese hamster ovary (CHO) lysates, while Lysate Mix 2 contains digests from human, yeast, mouse and Drosophila lysates. Samples were analyzed using either Zeno SWATH DIA or ZT Scan DIA, and database searches were performed using PEAKS Studio software v12 against a FASTA database combining canonical protein sequences from human, yeast, mouse, CHO and Drosophila. Venn diagrams indicate the number of overlapping proteins and peptides identified or those uniquely found using either Zeno SWATH DIA or ZT Scan DIA. ZT Scan DIA identified more unique peptides and proteins in both mixes than Zeno SWATH DIA.
Methods
Sample preparation: Yeast protein extract tryptic digest standard was purchased from Promega. Drosophila melanogaster protein lysate was purchased from GenLysate. Human HEK 293, MCF7, mouse 3T3, and Chinese hamster ovary (CHO) protein lysates were purchased from Innovative Research. Drosophila, HEK 293, MCF7, mouse, and CHO protein lysates were digested with trypsin and purified using S-Traps (Protifi) according to the manufacturer’s instructions. Two lysate digest mixtures were prepared (Table 1), both containing equivalent nanogram amounts of each specified digest; Lysate Mix 1 contained HEK 293, MCF7, yeast, mouse, and CHO digests, while Lysate Mix 2 contained HEK 293, MCF7, yeast, mouse, and Drosophila digests. Lysate Mixes 1 and 2 were each diluted to a final concentration of 100 ng/µL total protein in a buffer containing 5% acetonitrile and 0.1% formic acid in water.
Chromatography: Microflow LC separations were performed with a Waters M-Class UPLC system in direct-inject LC mode, using a Phenomenex XB-C18 analytical column (0.3 mm x 150 mm, 2.6 µm particle size), at a flow rate of 5 µL/min. Mobile phases consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). A 15-minute active gradient (25-minute total run time) was used for all experiments. Injection volumes were 2 µL for a total on-column loading of 200 ng digested protein per LC-MS run.
Mass spectrometry: Data-independent acquisition experiments were performed on a ZenoTOF 7600+ system using the OptiFlow Turbo V ion source with the microflow probe (1-10 µL/min electrode). Zeno SWATH DIA experiments used 65 variable-width windows spanning the TOF MS mass range 400-900 Da, MS/MS mass range 140-1750 Da, with MS/MS accumulation times of 13 msec, and Zeno trap pulsing turned on. ZT Scan DIA experiments used an estimated peak width at half-height (PWHH) setting of ≤ 1 sec. MS source parameters were as previously described [5]. Data was acquired in triplicate for all conditions.
Table 1: Description of the protein digest mixtures prepared for this study.
Table 2: FASTA protein sequence databases were constructed using canonical or canonical + isoform protein sequences for each organism. The number of protein sequences is indicated for each database.
Data processing: All data were processed using PEAKS Studio software version 12. Data acquired on Lysate Mixes 1 and 2 were processed using FASTA databases made up of combined protein sequences from Homo sapiens, Saccharomyces cerevisiae, Mus musculus, Cricetulus griseus, and Drosophila melanogaster. Two different FASTA database versions were used; one containing only canonical protein sequences and another containing canonical and isoform protein sequences. Table 2 shows the total number of proteins present for each organism in the two FASTA databases. The parameters used for the database search are shown in Table 3. Triplicate data files from the same experiment (specified Lysate Mix and MS method) were processed with each FASTA database.
Table 3: PEAKS Studio software database search settings.
Improved protein identification on complex digests using ZT Scan DIA
Using 15-minute microflow gradients (25-minute total run time), two different mixtures of lysate digests from various organisms were analyzed using either Zeno SWATH DIA or ZT Scan DIA on the ZenoTOF 7600+ system. These lysate mixtures were designed to challenge the system regarding the specificity of detections, as each differed in composition by one organism (CHO versus Drosophila). Replicate injections were made for each mix, with 200 ng total protein digest loaded on column per run. The ZT Scan DIA method (PWHH ≤ 1 sec) was compared with the 65 variable-window Zeno SWATH DIA method in terms of the resulting identifications of total proteins, protein groups, peptides and precursors after database searches using PEAKS Studio software. Table 4 summarizes the identification results for each mixture. The table shows 6.6% to 9.5% gains in total protein groups, 3.7% to 6.8% gains in total proteins, 10.8% to 14.8% gains in total precursors and 9.5% to 13.5% gains in total peptides using ZT Scan DIA over Zeno SWATH DIA.
Table 4: Total protein groups, proteins, precursors and peptides identified in each Lysate Mix sample using Zeno SWATH DIA or ZT Scan DIA. The percentage gains for each attribute with ZT Scan DIA are summarized. Database searches were done using the FASTA database made up of canonical protein sequences.
ZT Scan DIA identifies more unique peptides and improves protein sequence coverage
Using the built-in tools within PEAKS Studio software, the total proteins and peptides identified in each Lysate Mix sample were compared for the ZT Scan DIA and Zeno SWATH DIA results. Figure 1 shows Venn diagrams comparing the overlap in proteins and peptides. The Venn diagrams show a high degree of overlap in identifications, although approximately 1.5 to 2-fold more unique proteins and peptides were identified using ZT Scan DIA compared to Zeno SWATH DIA.
To dig deeper into the differences in identifications between the two methods, sequence coverage per protein was analyzed for each Lysate Mix, comparing ZT Scan DIA versus Zeno SWATH DIA. Figure 2 shows plots of the difference in protein sequence coverage per protein as a function of protein score (for proteins identified in both the ZT Scan DIA and Zeno SWATH DIA datasets) for Lysate Mix 1 (A) and Lysate Mix 2 (B). The plot shows that although some identified proteins had higher sequence coverage using Zeno SWATH DIA, a greater number of proteins were identified with higher sequence coverage using ZT Scan DIA. Differences in sequence coverage of as much as 35% were observed.
Figure 2. ZT Scan DIA results in higher protein sequence coverage than Zeno SWATH DIA. For each Lysate Mix sample, the difference in raw sequence coverage per protein [i.e., sequence coverage ZT Scan DIA minus sequence coverage Zeno SWATH DIA] is plotted as a function of protein score. Only proteins identified using both Zeno SWATH DIA and ZT Scan DIA are plotted.
Figure 3 summarizes the sequence coverage gains (i.e., proportion gain in % sequence coverage) using ZT Scan DIA, as a function of protein score, for both Lysate Mixes. These graphs display similar trends for both Lysate Mixes, highlighting that proteins with lower protein scores tended to show higher proportion gains in sequence coverage (as much as 700%, or 7-fold) using ZT Scan DIA. Since protein score can be correlated to the relative levels/abundances of proteins (and their related peptides), these results correspond well to earlier observations that ZT Scan DIA enhances the detection/identification of lower-level proteins and peptides [6]. The fact that protein sequence coverage was improved for lower-intensity proteins is significant, as low-abundance proteins are often the most challenging to detect and quantify in disease biomarker research.
Figure 3. Proportional increase in protein sequence coverage with ZT Scan DIA compared to Zeno SWATH DIA. For each Lysate Mix sample, the sequence coverage differential per protein [i.e. (sequence coverage ZT Scan DIA] – (sequence coverage Zeno SWATH DIA) ÷ (sequence coverage Zeno SWATH DIA)] is plotted as a function of protein score. Only proteins identified using both Zeno SWATH DIA and ZT Scan DIA are plotted. ZT Scan DIA generated gains of as much as 700% (i.e., 7-fold). More significant gains are observed for proteins with lower scores, relating to lower overall protein abundance levels.
ZT Scan DIA reduces the false discovery rate (FDR) of identifications
FDR was assessed for the datasets by comparing the database searches done against either the FASTA containing only the combined canonical protein sequences for the 5 organisms (48,408 total proteins) or the FASTA containing both canonical and isoform protein sequences (202,353 total proteins). Since Lysate Mix 1 contains CHO protein digest but not Drosophila melanogaster protein digest, searching that dataset would be expected to identify CHO proteins but not Drosophila proteins.
Conversely, Lysate Mix 2 contains Drosophila protein digest but not CHO protein digest; therefore, searching that dataset would be expected to produce identifications for Drosophila but not CHO. The proportion of “false” proteins in the FASTA database was easily predicted for each Lysate Mix, given that the total number of proteins present for each organism in the FASTA is known. The proportion of possible “false” proteins changes significantly when including isoforms compared to the FASTA containing only canonical sequences (see Table 5). The protein and peptide identification results for each Lysate Mix with each MS method were filtered to those reported with an expected FDR of 1%, and the number of peptides uniquely identified for CHO and Drosophila were compared. This data is also summarized in Table 5. The observed FDR rates for each Lysate Mix, run with either ZT Scan DIA or Zeno SWATH DIA, were plotted as a function of the proportion of possible false proteins in the database (Figure 4). These results clearly show that ZT Scan DIA reduces FDR for identifications from 0.03% to 0.02%, 0.28% to 0.26%, 0.49% to 0.38%, and 0.68% to 0.54%, translating to reductions in FDR by as much as 20%. Presumably, this enhancement results from the higher selectivity from the scanning quadrupole, generating MS/MS spectra of higher quality and specificity.
These results collectively highlight the unique benefits of ZT Scan DIA for proteomics research. These enhancements translate to higher protein sequence coverage and more confident identifications, which are crucial to users' understanding of complex biological processes and mechanisms involved in disease.
Figure 4. ZT Scan DIA reduces the false discovery rate (FDR) of identifications. The peptide FDR for Zeno SWATH DIA and ZT Scan DIA are plotted as a function of the proportion of false proteins in the database used for the database search (see Table 5). The FASTA database made up of canonical + isoform protein sequences contains more false proteins for a given Lysate Mix sample than the FASTA database made up of only canonical protein sequences. As the database size increases and contains potentially more false proteins, ZT Scan DIA is shown to further reduce the FDR of identifications relative to Zeno SWATH DIA.
Table 5: The proportion of false proteins in each FASTA database for a given Lysate Mix is shown, depending on the composition of the FASTA database used for the database search (containing either canonical only or canonical + isoform sequences). The observed peptide false discovery rate (FDR) is shown for each case based on either Zeno SWATH DIA or ZT Scan DIA results.
